我有一个相当复杂的查询,我正试图转换为使用Hive。
具体来说,我将它作为aws emr集群中的一个配置单元“步骤”来运行。
我已经试着为帖子清理了一些问题,只留下问题的本质。
完整的错误消息是:
FAILED: SemanticException [Error 10128]: Line XX:XX Not yet supported place for UDAF 'COUNT'
行号指向 COUNT
在select语句的底部:
INSERT INTO db.new_table (
new_column1,
new_column2,
new_column3,
... ,
new_column20
)
SELECT MD5(COALESCE(TBL1.col1," ")||"_"||COALESCE(new_column5," ")||"_"||...) AS
new_col1,
TBL1.col2,
TBL1.col3,
TBL1.col3 AS new_column3,
TBL1.col4,
CASE
WHEN TBL1.col5 = …
ELSE “some value”
END AS new_column5,
TBL1.col6,
TBL1.col7,
TBL1.col8,
CASE
WHEN TBL1.col9 = …
ELSE "some value"
END AS new_column9,
CASE
WHEN TBL1.col10 = …
ELSE "value"
END AS new_column10,
TBL1.col11,
"value" AS new_column12,
TBL2.col1,
TBL2.col2,
from_unixtime(…) AS new_column13,
CAST(…) AS new_column14,
CAST(…) AS new_column15,
CAST(…) AS new_column16,
COUNT(DISTINCT TBL1.col17) AS new_column17
FROM db.table1 TBL1
LEFT JOIN
db.table2 TBL2
ON TBL1.col311 = TBL2.col311
WHERE TBL1.col14 BETWEEN "low" AND "high"
AND TBL1.col44 = "Y"
AND TBL1.col55 = "N"
GROUP BY 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20;
如果我漏掉太多,请告诉我。
谢谢你的帮助!
更新
事实证明,我确实漏掉了太多的信息。对不起那些已经试图帮助。。。
我做了上面的更新。
按列删除第20组,例如:
GROUP BY 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19;
产生: Expression not in GROUP BY key '' ''
最新的
按列删除第20组并添加第一组,例如:
GROUP BY 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19;
产生:
Line XX:XX Invalid table alias or column reference 'new_column5':(possible column
names are: TBL1.col1, TBL1.col2, (looks like all columns of TBL1),
TBL2.col1, TBL2.col2, TBL2.col311)
第#行引用带有select语句的行。错误输出中只列出了tbl2中的这三列。
错误似乎指向 COALESCE(new_column5)
. 请注意,我有一个 CASE
在tbl 1 select中的语句,我用它运行 AS new_column5
.
1条答案
按热度按时间iq0todco1#
您正在寻址计算列名
new_column5
在计算它的同一子查询级别。这在 hive 里是不可能的。将其替换为计算本身或使用上层子查询。这是:
而不是这样: